Osteoarthritis (OA) is the most common disease of mankind. It is the most common cause of disability in this country and has enormous socioeconomic impact. OA is a disease of articular cartilage degeneration. Non- steroidal anti-inflammatory drugs (NSAIDs) are the most popular agents used in clinical practice to treat the joint pain and inflammation that is associated with OA. NSAIDs block the production of the enzyme cyclooxygenase resulting in a decreased production -of inflammatory prostaglandins, PGE2. Although NSAIDs are useful for pain management, recent studies have begun to produce data that require clinicians to re- evaluate our treatment strategy for OA. For example, when indomethacin was used for patients with hip OA, they required joint replacements in half the time and radiographic progression was twice as fast as when patients were treated with a weak PGE2 inhibitor, azapropazone. Other studies have not found NSAIDs to be better than acetaminophen for the treatment of painful knee OA. This relative lack of efficacy, and possibility of accelerated disease progression, coupled with the known gastrointestinal risks of these medications, especially to the elderly, have led us to re-evaluate NSAIDs as the first line medical therapy for osteoarthritis. Our dominant NSAID based approach to this disease may be resulting in unnecessary costs, unnecessary toxicity, and accelerated disability. These data allow us to hypothesize that NSAIDs, by inhibiting pain and inflammation in OA joints, may cause or encourage patients with OA to overuse damaged joints, resulting in accelerated joint degeneration and joint replacements at an earlier time or, alternatively, that treatment with NSAIDs may act to accelerate joint damage by altering cartilage metabolism and inhibiting joint healing. We further hypothesize that anti-inflammatory therapy with NSAIDs results in toxicities that lead to increased comorbidity and higher medical care utilization compared to analgesic use for OA. The specific aims of our proposed study are (1) to determine if non- steroidal anti-inflammatory drug therapy of NSAIDs accelerates joint degeneration compared to analgesic medications. (2) to determine if non- steroidal anti-inflammatory drug therapy results in greater comorbidity and higher medical care costs and utilization compared to simple analgesic medication. To accomplish these specific aims, we will randomize 200 knee and 200 hip OA subjects, defined by a Kellgren and Lawrence x-ray grade of 2 or 3, currently on NSAIDs, to either NSAIDs at their current dose or acetaminophen up to 4000 mg/day, for a period of four years. Primary outcome measures will be the rate of radiographic progression, pain and disability in the two groups. Secondary outcome variables will include medical care utilization, time to joint replacements, and medication side effect profiles. We will separately- identify and describe those clinical, demographic, and radiographic variables which predict accelerated progression in each group by multivariate analyses. By these methods, we will determine the long-term outcome of NSAID therapy versus analgesic therapy for the treatment of clinical OA of the knee and hip. This information is critical to improving the outcome of the disease that is the principal cause of disability in the elderly.